(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Children's Library | Biodiversity Heritage Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
See other formats

Full text of "Surface Potentials and Layer Charge Distributions in Few-Layer Graphene Films"




- 1—1 



Surface Potentials and Layer Charge 
Distributions in Few-Layer Graphene Films 

Sujit S. Datta 1 , Douglas R. Strachan 1 ' 2 , E. J. Mele 1 , A. T. Charlie Johnson 1 * 

CNJ ■ 1 - Department of Physics and Astronomy, 

2 - Department of Materials Science and Engineering, 
University of Pennsylvania, Philadelphia PA 19104 

Revised Manuscript Submitted: June 20, 2008; 


Graphene-derived nanomaterials are emerging as ideal candidates for postsilicon electronics. Eluci- 
dating the electronic interaction between an insulating substrate and few-layer graphene (FLG) films 
is crucial for device applications. Here we report electrostatic force microscopy (EFM) measurements 
revealing that the FLG surface potential increases with film thickness, approaching a "bulk" value for 
samples with five or more graphene layers. This behavior is in sharp contrast with that expected for 
conventional conducting or semiconducting films, and derives from unique aspects of charge screening 
by graphenes relativistic low energy carriers. EFM measurements resolve previously unseen electronic 
perturbations extended along crystallographic directions of structurally disordered FLGs, likely resulting 
from long-range atomic defects. These results have important implications for graphene nanoelectronics 

q ' and provide a powerful framework by which key properties can be further investigated. 

JZL • Graphene-derived nanomaterials are a promising family of structures for application as atomically thin tran- 

ly-v sistors, sensors, and other nanoelectronic devices. 1-3 Graphene, a honeycomb sheet of sp 2 -bonded carbon 

V^ . atoms, and nanotubes, graphene sheets rolled into seamless molecular cylinders, share a set of remarkable 

electronic properties making them ideal for use in nanoelectronics: tunable carrier type and density, ex- 
ceptionally high carrier mobility, and structural control of their electronic band structures. A significant 
advantage of graphene is its two-dimensionality, making it compatible with existing planar device architec- 
tures. This application requires fabrication processes where few-layer graphene (FLG) films are supported 
on insulating substrates. While prior theoretical work considered the effect of a substrate on the electronic 
structure of FLG, 7 " 12 few experiments have directly probed the graphene-substrate interaction. 10,13 Quan- 
titative understanding of charge exchange at the interface and the spatial distribution of the resulting charge 
carriers is a critical input to device design. 

Here we use electrostatic force microscopy (EFM) to probe the electrostatic interactions within FLG 
samples on Si02 substrates. We observe the effects of charge exchange between the FLG and the substrate 
on the surface potential, and we use its variation with thickness to quantify the layer charge distribution in 
the FLG. For films less than five layers thick, we observe substantial reduction in the electrostatic poten- 
tial measured at the exterior graphite surface as the number of layers decreases. We infer that the charge 
distribution induced in the FLG differs significantly from that expected for the conventional situation of a 
conducting or semiconducting thin film. We propose a nonlinear Thomas-Fermi theory for the FLG charge 

* E-mail: cjohnsonOphysics .upenn. edu 

carriers and find excellent quantitative agreement with the data. The theory predicts that as the surface 
potential asymptotically approaches the thick film value, it is reduced by a universal finite thickness correc- 
tion; the analytic form of this correction and its magnitude are in excellent accord with our measurements. 
These results highlight the importance of charge exchange between the FLG and the substrate, and provide 
a general framework for interpreting such interactions in this important family of vertically integrated struc- 

Our samples consist of FLG sheets transferred onto 300 nm SiO^/Si by mechanical exfoliation under 
ambient. 14 A Vceco Dimension 3100 atomic force microscope (AFM) with a conductive tip is used to record 
the sample topography and EFM phase in a two-pass mode. In the first pass, the AFM tip traces the sample 
topography (figures 1 and 2A). In the second (interleave) pass along the same scan line, the tip is biased with 
a DC voltage Vu p and retraces the topography at a fixed lift height h above the surface. The cantilever is 
mechanically driven on resonance, and the phase shift of the cantilever oscillation is measured as a function 
of tip position. Modeling the cantilever as a harmonic oscillator of resonant frequency lvq, spring constant 
k, and quality factor Q, and adopting the standard convention that the measured phase shift $ = <fi + 7r/2 
(where <j> is the phase shift between the driving force and the cantilever oscillation), the phase shift over the 
sample, due to tip-sample capacitive coupling, is: 15 ' 16 

$(x,y) = ~±C"{h) (V tip - V s {x,y)f (1) 

where C"(h) is the second derivative of the difference between the tip-sample capacitance and the tip-bare 
substrate capacitance as a function of h, and V s (x,y) is the local electrostatic potential on the sample sur- 
face. The phase shift of equation (1) is zero when Vu p equals the value of V s directly below the tip, so the 
surface potential can be mapped by EFM with spatial resolution of order 20 nm. EFM is an established 
technique for mapping the surface potential of planar samples, including semiconductor nanocrystals, thin 
films of polymers and polymer blends, and self-assembled monolayers. 17 ~ 21 

From the measured sample topography, we identify well-defined FLG regions with numbers of graphenc 
layers n ranging from 2 to 18. Figure 2A is a typical topographic image, and figures 2B-C the corresponding 
EFM phase images for two different tip voltages. The phase shift A$ of the FLG regions with respect to 
the bare Si02 substrate is always negative. FLG regions with different numbers of graphene layers exhibit 
different values of A$ (figure 2d); fluctuations in A$ within each bulk FLG region are indistinguishable from 
instrumental noise. The relative contrast between FLG regions is reversed in images taken with tip voltages 
of opposite polarity, indicative of different values of V s (see below). The data indicate that the surfaces of 
each FLG region are equipotcntials, with surface potential that varies with n. 

We quantify V s by measuring A<& as a function of Vu p , as shown in figure 3 A for FLGs that are two and 
five layers thick. The maxima of the parabolic fits to the data (see equation (1)) lie at the different values 
of V s . Figure 3B shows the variation in A V s = V s — V™ ax as a function of thickness for FLGs with layer 
number ranging from 2 to 18. The surface potential is always positive, indicating hole doping in ambient. 1 
For FLGs thicker than 5 layers (~1.7 nm), the surface potential is approximately constant (V smax — 779 
mV), and V s decreases sharply for thinner FLGs. 

The central result of our EFM experiments is that the surface potential above FLG films varies strongly 
(over hundreds of mcV) and monotonically with the film thickness. In the following paragraphs, we model 
the data quantitatively by assuming that the variation in surface potential is determined by the spatial distri- 
bution of doped charges (holes) that are transferred to the graphene from a thin (< 1 nm) interfacial layer of 
traps or defects at the silica surface. Based on measured current-backgate voltage characteristics of exfoliated 
FLG conducted in our lab and elsewhere, 1_3 we expect an areal carrier number density o$ ~ 10 12 cm~ 2 . With 
these assumptions, we can use the thickness dependence of V s to probe the layer-by-layer charge distribution 
in the FLG. We find that this model of intrinsic screening in the FLG provides an excellent quantitative 
description of our observations. This is consistent with the expectation that extrinsic contributions to the 

surface potential (e.g., charge contamination and adsorbed water) typically lead to a uniform shift in surface 
potential that does not vary with FLG thickness. 

The graphene charge distribution is determined by competition between kinetic energy of the doped 
carriers and their interaction with a self-consistent electrostatic potential. We describe it with a nonlinear 
Thomas-Fermi (TF) theory for the doped carriers in the continuum limit, appropriate when the layer areal 
carrier density a(z) changes smoothly on a scale of the interlayer spacing d, assumed to be much less than 
the FLG thickness D. In the theory, the kinetic energy of the doped carriers in each layer increases with 
areal number density as a(z) 3 ' 2 instead of a{z) 2 as it would for a conventional compressible (conducting) 
system. This crucial difference reflects the massless low energy graphene dispersion relation for doped car- 
riers used in the non-linear TF theory, which does not treat the effect of quantum coherence between the 
graphene layers. The assumption of decoupled graphene layers is accurate when the local Fermi energies 
are far from the charge neutrality point and their shifts are large compared to the size of the interlayer 
hopping amplitudes. We also note that this was the model successfully applied in references 22 and 23 to 
study intercalated graphite and has been found 24 to have good agreement with a self-consistent tight binding 
model in the range of layer charge densities relevant to our FLG samples. 

The theory is formulated in terms of a quantity f(z) = ^/ a{z) which is a solution to the nonlinear 

d 2 f(z) _ 2~p 2 

dz 2 d 

t(z) (2) 

where 7 = 2^/tt1%vf/3 1 (3 — Aire 2 /3j, subject to the boundary conditions [df/dz] z =o = 2/3<7o and [df/dz] z =D = 
(see supporting text for details). To understand the solutions we scale f(z) by its maximum value /o at 
the interface and define the ratio parameter rjj = f(D)/fo by the integral relation 


{l - r ° )1 " 1 7^r 2 l^r= r (3) 

,■'■• i 1 ! - / "'" - 9 ( ^oD^\ 

The dimcnsionlcss parameter T determines rjj and consequently the nature of the FLG screening. When 
r > 1 the system is in a strong coupling regime dominated by its electrostatic energy; here rjj < 1 and the 
doped hole distribution is strongly peaked near the oxide interface. The FLG contribution to the measured 
V s can be expressed in terms of ro , giving the central result of the Thomas-Fermi theory 

V s = ^ ( 3^)V3_l_^_ + Fo (4) 

where Vo = 3"f(3aod s /2 is a constant offset due to the charge distribution in the substrate. Here d s is the 
characteristic depth of ionized acceptors in the Si02, estimated to be 0.5 — lnm if the charge resides in 
an interfacial defect or impurity layer, as expected. Thus the graphene contribution dominates the surface 
potential whenever <r < 2d/(0 2 d 3 ) ~ 6 x 10 12 cm -2 . In this situation, variations in the surface potential V s 
with film thickness D are controlled entirely by the distribution of charge in the FLG in the experimentally 
relevant situation where cto is independent of D. 13 

Our samples are described by the strong coupling limit of the TF model, for which wc find V s w 
(37/2)(3/3d<T 2 ) 1 / 3 — 9~/d/2f3D 2 . This explains the observed asymptotic behavior of V s , with the thick film 
value (V™^) determined by the areal charge density. As it approaches this limit, V s contains a universal 
(independent of <7o) finite size "thin film" correction. Using the accepted values 7 = 6.5eV-A and j3 = 9.5, 
we calculate the finite size correction 

V s max -V S = AV S = -W.3K(V-A 2 )/D 2 (5) 

including the effect of a static dielectric constant n perpendicular to the layers. Although this 1/D 2 de- 
pendence resembles the confinement energy of a quantum mechanical gas of massive doped holes, here it 
arises from a competition between the two-dimensional kinetic energy for a gas of massless carriers (which 
suppresses spatial variations in cr(z)), and the Coulomb interactions in the third dimension (which favors 
them). Appropriate values of the dielectric constant for the FLG/Si02 interface region lie in the range 2 < k 
< 3; 22-24 the solid lines in figure 3 correspond to these lower and upper bounds, while the dotted line (using 
K = 2.5) gives an excellent fit to the data. Though our present measurements do not exclude the possibility 
of an exponential decay (as expected for screening by massive carriers) , the excellent quantitative agreement 
with eq 5 gives strong support for the non-linear Thomas-Fermi behavior along the c-axis in FLG samples. 

Solving equation (2) with d s = 0.5nm and VJ nax = 779mV gives oo = 6 x 10 12 cm -2 , which is understood 
as arising from charge exchange to a small acceptor density at the interface. This indicates that for our 
samples, the potential drop in the substrate and the FLG contribute roughly equally to the overall surface 
potential. Furthermore, it confirms that our FLG films realize the strong coupling limit of the model (r > 1) 
where the doped holes accumulate close to the interface with the oxide, with a strong power law decay into 
the FLG film. The rapid nature of this decay can be inferred directly from the rapid convergence of the 
surface potential V s to its thick film limit (figure 3B). 

Our data is consistent with recent angle-resolved photoemission spectroscopy measurements of FLG on 
SiC; 13 indeed, neither study finds any evidence for the oscillatory charge and potential profile that are pre- 
dicted from a linear response theory of the screening. 25 It is interesting to note that our analysis predicts 
that the induced charge profile in the graphene film has a power law decay, a situation that is thought to 
describe the layer charge distributions found in graphite intercalation compounds. 22-24 Furthermore, our 
model of charge exchange to a small density of defect/interface states is quite similar to conclusions reached 
via qualitative considerations of Fermi energy equilibration for pentaccne films on Si02- 26 

These results suggest that EFM might also be used to investigate electronic perturbations in FLG films. 
In particular, structural disorder is expected to strongly alter FLG electronic properties 27 ' 28 in ways yet to 
be understood. To this end, we extend EFM measurements to the study of disordered FLGs, specifically a 
representative sample consisting of 9 graphene layers in the bulk (figure 4A) . The presence of well-defined 
folds suggests a high density of defects and dislocations in the FLG, and associated local stresses within 
it. 29,30 A comparison to the schematic in figure 4C shows that the many folds along the right edge of the 
film all fall along (100) "zig-zag" and (100) "armchair" directions of the graphene lattice. 

Although mechanical stresses and folding along crystallographic directions of graphitic samples have 
been studied, 29,30 they have not been correlated with electronic properties. Strikingly, EFM phase images 
of the disordered FLG film indicate parallel stripe-like electronic perturbations along specific directions of 
the underlying honeycomb lattice (figure 4B). Topographical features corresponding to these stripes are not 
discernable by AFM. For clarity, the ends of one such stripe are indicated in figure 4B with arrows, where 
the potential over the stripe is reduced by ~0.25V compared to the unperturbed bulk region, where we 
find V s = 0.76V, in agreement with figure 3. Remarkably, all observed electronic stripe perturbations lie 
parallel (±3°) to a single fold direction of the underlying graphene lattice. Although we do not know the pre- 
cise origin of these electronic perturbations, there are several possible sources, including grain boundaries or 
microcracks, 26 'scars', 25 or sp 3 -like rehybridization defects. 29,30 Interestingly, similar electronic perturbations 
were observed at surface defects such as line dislocations and grain boundaries in pentacene multilayers. 26 
The data suggest the intriguing possibility that EFM could prove useful for nanoscalc imaging of local defects 
and their impact on electronic structure in graphene-based devices. 


We thank Y. Zhao (P. Kim group, Columbia University) for help with sample preparation. Research sup- 
ported by the Nano/Bio Interface Center through the National Science Foundation NSEC DMR-0425780; 
the JSTO DTRA, the Army Research Office Grant W911NF-06-1-0462; the Department of Energy Grant 
DE-FG02-06ER45118; and the Intelligence Community Postdoctoral Fellowship Program. 


1. Novoselov, K. S. et al. Electric Field Effect in Atomically Thin Carbon Films. Science 306, 666 - 669 

2. Schcdin, F. et al. Detection of individual gas molecules adsorbed on graphene. Nature Materials, 
Advanced Online Publication (2007). 

3. Williams, J. R., DiCarlo, L., Marcus, C. M. Quantum Hall Effect in a Gate-Controlled p-n Junction 
of Graphene. Science 317, 638-341 (2007). 

4. Novoselov, K. S. et al. Two-Dimensional Gas of Massless Dirac Fermions in Graphene. Nature 438, 
197-200 (2005). 

5. Ohta, T., Bostwick, A., Seyllcr, T., Horn, K., Rotenbcrg, E. Controlling the Electronic Structure of 
Bilayer Graphene. Science 313, 951 - 954 (2006). 

6. Zhang, Y., Tan, Y.-W., Stormer, H. L., Kim, P. Experimental observation of the quantum Hall effect 
and Berry's phase in graphene. Nature 438, 201 - 204 (2005). 

7. Guinea, F. Charge Distribution and Screening in Layered Graphene Systems. Physical Review B 75, 
235433 (2007). 

8. McCann, E. Asymmetry gap in the electronic band structure of bilayer graphene. Physical Review B 
74, 161403(R) (2006). 

9. Min, H., Sahu, B., Banerjee, S. K., MacDonald, A. H. Ab Initio Theory of Gate Induced Gaps in 
Graphene Bilaycrs. Physical Review B 75, 155115 (2007). 

10. Zhou, S. Y. et al. Substrate-induced bandgap opening in epitaxial graphene. Nature Materials 6, 
770-775 (2007). 

11. Adam, S., Hwang, E. H., Galitski, V. M., Sarma, S. D. A self-consistent theory for graphene transport. 
Proc. Natl. Acad. Sci. 104, 18392 (2007). 

12. Katsnclson, M. I., Gcim, A. K. Electron scattering on microscopic corrugations in graphene. Phil. 
Trans. R. Soc. A 366, 195 (2008). 

13. Ohta, T. et al. Interlayer Interaction and Electronic Screening in Multilayer Graphene Investigated 
with Angle- Resolved Photoemission Spectroscopy. Physical Review Letters 98, 206802 (2007). 

14. Information on materials and methods is available online. 

15. Staii, C, Johnson, A. T., Pinto, N. J. Quantitative Analysis of Scanning Conductance Microscopy. 
Nano Letters 4, 859 - 862 (2004). 

16. Bockrath, M. et al. Scanned Conductance Microscopy of Carbon Nanotubes and A-DNA. Nano Letters 
2, 187-190 (2002). 

17. Coffey, D. C, Ginger, D. S. Time-Resolved Electrostatic Force Microscopy of Polymer Solar Cells. 
Nature Materials 5, 735 - 740 (2006). 

18. Lei, C. H., Das, A., Elliott, M., Macdonald, J. E. Quantitative electrostatic force microscopy phase 
measurements. Nanotechnology 15, 627 (2004). 

19. Silveira, W. R., Muller, E. M., Ng, T. N., Dunlap, D. H., Marohn, J. A. in Scanning Probe Microscopy: 
Electrical and Electromechanical Phenomena at the Nanoscale (eds. Kalinin, S. V., Gruverman, A.) 
(Springer Verlag, New York, 2005). 

20. Takano, H., Wong, S.-S., Harnisch, J. A., Porter, M. D. Mapping the subsurface composition of organic 
films by electric force microscopy. Langmuir 16, 5231 (2000). 

21. Gordon, J. M., Baron, T. Amplitude-mode electrostatic force microscopy in UHV: Quantification of 
nanocrystal charge storage. Physical Review B 72, 165420 (2005). 

22. Pietronero, L., Strssler, S., Zeller, H. R., Rice, M. J. Charge Distribution in c Direction in Lamellar 
Graphite Acceptor Intercalation Compounds Physical Review Letters 41, 763-767 (1978). 

23. Safran, S. A., Hamann, D. R. Electrostatic interactions and staging in graphite intercalation com- 
pounds. Physical Review B 22, 606-612 (1980). 

24. Safran, S. A., Hamann, D. R. Self-consistent charge densities, band structures, and staging energies of 
graphite intercalation compounds. Physical Review B 23, 565-574 (1981). 

25. Guinea, F. Charge distribution and screening in layered graphene systems. Physical Review B 75, 
235433 (2007). 

26. Puntambekar, K., Dong, J., Haugstad, C, Frisbie, C. D. Structural and Electrostatic Complexity at a 
Pentacene/Insulator Interface. Advanced Functional Materials 16, 879 (2006). 

27. Morozov, S. V. et al. Strong suppression of weak localization in graphene. Physical Review Letters 97, 
016801 (2006). 

28. Peres, N. M. R., Guinea, F., Neto, A. H. C. Electronic Properties of Disordered Two-Dimensional 
Carbon. Physical Review B 73, 125411 (2006). 

29. Ebbesen, T. W., Hiura, H. Graphene in 3-Dimensions: Towards Graphite Origami. Advanced Materials 
7, 582 (1995). 

30. Li, L. X. et al. Tearing, folding and deformation of a carbon-carbon sp 2 -bondcd network. Carbon 44, 
1544 (2006). 

Figure 1: Schematic of the experiment. FLG samples of different thicknesses (here, 9 and 18 layers) are 
deposited on a SiC>2 substrate on a highly-doped Si ground plane. Sample topography and EFM phase are 
imaged simultaneously. Orange surface plot is 4.1 /im x 4.1 /iin AFM topography image, with low-pass 

0.5 1.0 

Lateral Position (nm) 

Figure 2: AFM and EFM of few-layer graphene. (A) AFM image of FLGs on SiC^/Si substrate; color scale 
is lOnm. (B, C) EFM phase images of the sample, with Vu p = -2V and +3V, respectively. Color scales are 
5.0°. (D) Average of 30 line scans of topography and phase centered along the dashed lines in AFM and 
EFM images, with low-pass filtering. Black curve corresponds to (A), red curve to (B), and blue curve to 
(C). Scale bar in each image is 1.5 /im. 


»_„^ • 2 layers. 

//< ^sX, " ^ layers 


« -1 

rf \\ 



*- 9 

// \ \ 


11 \ \ 

o ? 


< -4 

// \ \ " 

v m W 




5 10 15 



Figure 3: Surface potential measurement by EFM. (A) EFM phase shift versus tip voltage for FLGs with 
2 and 5 layers (black and red data, respectively). Solid lines are parabolic fits to the data. Error bars 
represent instrumental noise. (B) Plot of the variation in surface potential, AV S = V s — V™ 11 *, versus FLG 
thickness D, normalized by single graphene layer thickness (0.34nm). Data points for £>/0.34nm w 2, 3, 
7, 8 are averages of data from two different regions of equal D, while other data points are over a single 
region of the specified thickness. Error bars represent statistical uncertainty in the value of V s obtained from 
parabolic fits as in Fig 3A. Curves represent predictions of a nonlinear Thomas-Fermi screening model with 
background dielectric constant k — 2 (red), 2.5 (dashed), and 3 (blue). Inset: AV S as a function of 1/D 2 , 
with the power-law fit predicted by the model. 

Figure 4: Extended defects in few-layer graphene are revealed in EFM. (A) AFM topographic image of 
a disordered FLG sample; color scale is lOnm. (B) Schematic of a possible structure, indicating folding, 
symmetry axes, and FLG crystallographic axes. (C) Corresponding EFM phase image, with Vu p = +1.5V. 
Color scale is 1.0°. All scale bars are 1.5 fim. 



Supporting Information: 

Surface Potentials and Layer Charge 
Distributions in Few-Layer Graphene Films 


Sujit S. Datta 1 , Douglas R. Strachan 1 ' 2 , E. J. Mele 1 , A. T. Charlie Johnson 1 * 

1 - Department of Physics and Astronomy, 

2 - Department of Materials Science and Engineering, 

University of Pennsylvania, Philadelphia PA 19104 


1 Materials and Methods 

"£i \ 1.1 Graphene Preparation 

Graphene samples are synthesized using a mechanical exfoliation technique similar to that described in Ref. 
1. We start with a highly-doped Si substrate with 300nm of thermally-grown Si02, cleaned with acetone 
and isopropyl alcohol. Adhesive scotch tape is used to extract and exfoliate a starting piece of bulk Kish 
graphite (Toshiba Ceramics, San Jose, CA). After the graphite has been sufficiently thinned, the tape is 
pressed against the Si02 surface and gently rubbed with the back of a tweezer for approximately 15s. The 
strength with which the tape is rubbed is varied across the sample surface, resulting in distinct FLG regions, 
some structurally pristine and others disordered. FLG regions of varying thickness are selected by visual 
^ ' inspection with an optical microscope. 



1.2 Details of AFM and EFM measurements 

Two different kinds of metal-coated SPM tips are used for AFM and EFM measurements: Ti-Pt tips (NSC18, 
Mikromasch) are characterized by force constant k ~ 3.5N/m, quality factor Q ~ 250, resonant frequency 
wo ~ 75kHz, tip curvature radius ~ 40nm; Cr-Au tips (NSC15, Mikromasch) are characterized by k ~ 
40N/m, quality factor Q ~ 200, resonant frequency ujq ~ 325kHz, tip curvature radius ~ 50nm. AFM height 

measurements are done using intermittent contact mode. 

1.3 Identification of Number of Layers 

Previous experiments have indicated the possible presence of a "dead layer" between the FLG and the 
Si02 substrate that causes AFM measurements of the graphene thickness to be increased by up to several 
angstroms 1 . If a dead layer were present, accurate identification of the number of layers in the FLG would be 
complicated. However, the number of layers present in a FLG region can be accurately measured by AFM if 
the film contains folds or wrinkles. AFM measurements of a folded FLG region prepared on the same chip as 

* E-mail: cjohnsonOphysics .upenn. edu 

our samples indicate that the thickness of the dead layer, if there is one, is much smaller than the thickness 
of a single graphene layer (0.34 nm), so we are able to accurately determine the number of graphene layers 
in our FLG films. An example of this is shown in fig. SI. 

1.4 EFM Measurements 

By convention, the EFM phase shift A<I> is measured with respect to the bare substrate (i.e. A$ = over 
a SiC>2/Si region with no deposited FLG). Each data point of fig. 3A (main text) represents the average of 
~5-10 different line scans over the same region (or the average of 10-20 line scans taken over two different 
regions of identical thickness - see the caption of Figure 3). As explained in detail in Ref. 2, conducting 
samples exhibit a negative phase shift, while insulating samples exhibit a positive phase shift that depends 
on the sample dielectric constant. For electrically floating conducting samples such as FLGs, changes in 
4>{x, y) (given by equation 1 of the main text) result from changes in C"(ft), a geometric effect, or changes in 
the local sample surface electrostatic potential V s (x, y). Because the surface roughness of FLG bulk regions 
is of order 0.1 - 0.5 nm (^0.3% of the tip radius 3 ), C"(ft) is constant to an excellent approximation, and 
changes in A<J> reflect changes in V S . A 

We conducted multiple control experiments to verify the accuracy and reproducibility of the surface po- 
tential measurement. Height-dependent EFM measurements (fig. S2A) of A$ over the same FLG region 
at fixed V t ip have a power-law form A<I> ~ ft, -1 - 6 , characterized by an exponent between that expected for 
a cone-plane geometry (C"(h) ~ ft. -1 ) and a sphere-plane geometry (C"(ft) <~ ft -2 ), as seen by others 5 . 
EFM measurements of the same FLG region were taken using two different lift heights, with the A$-V t i p 
data shown in Fig S2B. The measured value of V s is verified to be independent of the lift height ft, as expected. 

2 Supporting Text 

2.1 Outline of Thomas- Fermi Theory 

The Thomas-Fermi (TF) theory describes the distribution of layer areal number densities a(z) that are in 
equilibrium with respect to fluctuations Sa(z). The TF theory does not treat the effect of quantum coherence 
between the layers, and it applies in the controlled limit that the interlayer tunneling is tuned to zero. It 
is accurate when the variations of the interlayer potential or equivalcntly, when the local shift of the Fermi 
energies are large compared to the size of the interlayer hopping amplitudes. 

For doped carriers described by the conical dispersion relation E{k) = hvpk, an excess areal carrier 
density contributes to the kinetic energy (per unit area) of each layer Ki — 2y / 7rftuF<7 i /3. The doped 
carriers in each layer also interact with a compensating density — eoo in the substrate and with the charges 
in each layer. In the continuum limit, the grand potential for this system is 

= ^1+^2 (f) 

Q 1= -^ (l(r{zf /2 - H<j(z) - 2ire 2 z<r (T(z)] (2) 

Oa = - I' \e 2 %%o{z)\z-z'\a{z') (3) 

./n a a 

where 7 = 2y/irhvF /3 and \i is the chemical potential. Minimizing fl and defining f(z) = ^fa[z) one finds 

*lf(z) = n + /3z<To + 2 I dz'\z - z'\a(z') (4) 

and differentiating twice with respect to the observer coordinate z we obtain the equation quoted in the text 


11 = W 

dz* d J[ ' 

with (3 — 47re 2 /37. The boundary conditions can be determined by considering the behavior of the first 
derivative of eq. (4), along with the constraint of charge conservation 


= /3Uo- / ^r<r(z')) -2/3ao 



= <(»» + r^r<r{z'))=o 


=£> \ JO « J 

The solutions of eq. (5) are then obtained by using the conservation law 

dz \ 2 I dz 

2/? ,3 




and then directly integrating df /dz to find f(z). In terms of a dimcnsionlcss variable of integration u 
f (z)//(0) we find 

(1-^) 1/6 


V u3 - 


( pa D 3 
I 3d 



where rjj = f(D)/f(0). Equation (9) provides a convenient relation between the dimcnsionlcss coupling 
parameter T and the screening parameter rp. Once rjj is determined (4) can be used to find the graphene 
contribution to the surface potential 




The first term on the right hand side is the contribution to the surface potential from the charge distribution 
in the graphene. V(0) is the potential at the graphene/silica interface and is offset from the potential deep 
in the silica because of the charge distribution in the charged acceptor layer. Setting the potential to zero 
in the silica bulk, we have 

V(0) = Y&ods - 2m 2 aod a 


for an areal acceptor density ctq within depth d s of the interface. 

Supporting References 

1. Novoselov, K. S. et al, Science 306, 666 - 669 (2004). 

2. Staii, C, Johnson, A. T., Pinto, N. J., Nano Letters 4, 859 - 862 (2004). 

3. Ishigami, M., Chen, J. H., Cullcn, W. G., Fuhrer, M. S., Williams, E. D., Nano Letters 7, 1643 (2007). 

4. We note that because our measurements are performed under ambient laboratory conditions, the 
measured V s is expected to be offset from the value measured under UHV. For example, although 
graphene is well known to be hydrophobic, we can not rule out the presence of a thin adsorbed water 
layer on the sample surface, or hydrogen bonded to the Si-OH silanol groups at the SiC>2. However, 
our investigations are of robust changes in the FLG surface electrostatic potential, and thus the slight 
offset introduced in our measurements by such effects does not affect our conclusions. Furthermore, 
we note that all the measurements presented are robust over a period of multiple weeks. 

5. Bclaidi, S. et al, J. Appl. Phys. 81, 1023 (1997). 





12 3 4 5 

Height (nm) 

Figure 1: AFM across FLG fold. Height histograms acquired across an unfolded FLG/substrate edge (top) 
and a folded FLG/substrate edge (bottom), for neighboring pristine regions of the sample shown in figure 
4 of the main text. The height difference between the bulk region and the Si02 is ~2.73 nm, while the 
height difference between the folded region and the Si02 is ^5.3 nm, almost exactly twice as large. We thus 
conclude that the unfolded region consists of 8 layers (~ 2.73 nm / 0.34 nm, 0.34 nm being the thickness of 
a single graphene layer), and that the thickness of the "dead layer" (if any) in this experiment is negligible 
compared to the interlayer spacing. 




e i 

1 1 ' 1 p p p 1 

• 5Gnm 
"« 30nm 



% ■ 






© -4 



40 60 80 100 


4 -2 



Lift Height h (r 


V (V) 

Figure 2: Results of EFM control experiments. (A) Negative of EFM phase shift (— A$) measured as a 
function of the lift height h, over three different FLG regions (different colors). The dashed line indicates the 
functional form — A$ ~ h~ 16 . (B) A$ versus tip voltage Vu p for the same FLG region of height ~6.1nm 
(18 layers), for lift heights h = 30nm and 50nm. The measured AV S = V s - V s max = 0.02 ± 0.06V agrees 
well with data in figure 3 of the main text, and confirms that the measured value of V s is not sensitive to 
the lift height h.